An RFI Mitigation Strategy for the Allen Telescope Array Geoffrey C. Bower UC Berkeley.

An RFI Mitigation Strategy for the Allen Telescope Array

Geoffrey C. Bower

UC Berkeley

The Allen Telescope Array

350 element interferometer

6.1 m offset Gregorians 0.5-11.2 GHz Science goals

– SETI– HI survey machine– Transient radio sources

Strategy Overview

Know the enemy Put many arrows in the quiver

– No Magic Bullets

Exploit large N capability Implement at all levels

Available Tools

Modeling of environment Measurement Real time strategies

RFI Modeling: Hat Creek Terrain

RFI Modeling: Free Space

3 x 3 degree region centered at Hat Creek Contours of path loss Visualyse simulation

RFI Modeling: ITU-R P.526

Model includes topography, diffraction, tropospheric scatter, ducting/layer reflection, fade

Future modeling: frequency, scale, model parameters, Longley-Rice

RFI Monitor

Simple design– Discone antenna– Broadband amp– Spectrum analyzer– Linux PC on ‘net

0.1 to 10 GHz– 0.3 to 3 GHz currently

Isotropic sensitivity

… (One of) the Enemies is Us

Is Glonass Present at 1612 MHz?

Tracking Glonass

…The Enemy is Us, Part 2

And Tracking GPS

Aircraft DME

RFI Sources Detected at Hat Creek

Self-interference Satellites --- GPS, Iridium, DARS Aircraft Radar TV, cell phone, pager Microwave ovens

Real Time RFI Mitigation

Monitoring Artificial horizons around known interferers

Time Blanking Adaptive canceling Interferometric Nulling Postcorrelation Analysis

Time Blanking: Radar

Time Blanking: DME

Time Blanking Implementation

Done in backends Correlator will have

– 100 sec precision– External & Internal Triggers– Channel, Antenna Configuration Undetermined

Adaptive Canceling

Rapid Prototyping Array

Cancellation of Glonass

Cancellation of GPS

ATA Implementation of Canceling

Multiple reference antennas (~10) Correlator/black box operating on phased-

array signal Wiener or LMS solutions

Interferometric Nulling: Two Slits

Interferometric Nulling

Nulls with the ATA

Muliple Nulls

Nulls in Frequency Space

The Wages of Nulling

Requires large-N to be effective

Loss in SNR with number of nulls

Must be able to update complex gains at ~100 Hz

Matrix inversion is computationally expensive

Postcorrelation Analysis

Examine output of correlator/visibility matrix ~100 element PC cluster Data rate:

– 350x349/2– 4 Stokes– 1024 Channels– 16 bit– 100 Hz sampling

400 Gbits/sec: too much!

How to Use the Tools

Blanking– Radar, Aircraft, intermittent signals

Adaptive Canceling– Reference antenna(s) on transmitter

Interferometric Nulling– Predictable trajectories: satellites, fixed transmitters

Postcorrelation Analysis– Broadly adaptable

Future Steps

Algorithm development More modeling of environment Implement canceller, external blanker Angle of arrival methods